HPV genotypes screening improve cervical cancer prediction

HPV genotypes screening improve cervical cancer prediction | Image Credit: © onephoto - © onephoto - stock.adobe.com.

According to a recent study published in JAMA Network Open, models for predicting cervical cancer among women with a diagnosis of high-risk human papillomavirus (hrHPV) are improved with the inclusion of human papillomavirus (HPV) genotypes.

There were approximately 604,000 new cases of cervical cancer and 342,000 deaths worldwide in 2020, making it the fourth most common cancer in women and a severe threat to women’s lives. A decrease in cervical cancer incidence and mortality has been associated with well-established screening programs, but implementing these programs is an issue in developing countries.

While screening programs have been implemented in China, the country has a coverage rate below 30%. Many countries are also facing a lack of medical resources and skilled health care personnel. This has made it difficult for cervical cancer screening programs to cover all women.

HrHPV is associated with an increased risk of cervical cancer, and different HPV genotypes are associated with different cervical cancer risks. However, few models consider including HPV genotypes.

Investigators conducted a study to develop a model for cervical cancer screening including HPV genotypes. A cervical cancer screening program occurred in Xiangyang City, China, from January 15, 2017, to February 28, 2018.

Participants included women aged 30 years or older with over 1 year of sexual activity and no pregnancy, prior HPV vaccination, or hysterectomy or pelvic radiation therapy history. All participants had a positive test for hrHPCV infection and were evaluated through a pelvic examination, HPV genotype testing, and questionnaires.

Data collected included participant demographic characteristics, medical history, menstrual status, sexual behavior factors, and family cancer history. Patients underwent a visual vulva inspection, internal vagina and cervix speculum examination, and bimanual adnexa and uterus palpation during pelvic examination.

Vaginal status was assessed through vaginal bacteria, miscellaneous bacteria, and number of white blood cells. Four categories were created, with categories 1 and 2 considered normal and categories 3 and 4 considered abnormal.

The Cobas 4800 HPV test (Roche Diagnostics) was used to detect HPV genotypes. Categories included HPV-16, HPV-18, other hrHPV genotypes, HPV-16 plus HPV-18, HPV-16 plus other hrHPV genotypes, HPV-18 plus other hrHPV genotypes, and HPV-16 plus HPV-18 plus other hrHPV genotypes.

Microorganism infection in the vaginal microenvironment was also considered during evaluation. The primary outcome of the study was cervical intraepithelial neoplasia 3 or worse (CIN3+), while the secondary outcome was cervical intraepithelial neoplasia 2 or worse (CIN2+).

There were 314,587 women who received cervical cancer screening, 7.8% of which had hrHPV and 11% of which were excluded because of dropout. The remaining women were placed into training data set or validation data set groups. Of the training data set group, 2.4% received a CIN3+ diagnosis and 4.6% a CIN2+ diagnosis, compared to 2.3% and 3.2% respectively in the validation data set group.

The highest set of genotypes observed were other hrHPV genotypes, seen in 77.2% of the training group. Other genotypes observed in this group included HPV-16 in 11.8%, HPV-16 plus other hrHPV genotypes in 5.2%, HPV-18 in 3.5%, HPV-18 plus other hrHPV genotypes in 1.6%, HPV-16 plus HPV-18 in 0.3%, and HPV-16 plus HPV-18 plus other hrHPV genotypes in 0.3%.

When predicting CIN3+, models saw an area under the receiver operating characteristic curve (AUROC) value improvement of 35.9% when including HPV genotypes. For CIN2+, the AUROC value improved by 41.7% when including HPV genotypes.

These results indicated improved cervical cancer prediction from models including HPV genotypes. Investigators concluded these models may allow for early cervical cancer diagnoses in low-resource settings.

Reference

Xiao T, Wang C, Yang M, et al. Use of virus genotypes in machine learning diagnostic prediction models for cervical cancer in women with high-risk human papillomavirus infection. JAMA Netw Open. 2023;6(8):e2326890. doi:10.1001/jamanetworkopen.2023.26890